基于局部特性实现单帧图像小目标检测的研究

Study of Small Target Detection in Single Frame Image Based on Local Characteristics

以目标成像点的扩散理论为基础,建立小目标在空域上的灰度特性模型,分析目标、背景和噪声的基本特性。由形态学开闭运算得到各像素位置的灰度变化值,再根据此值确定潜在目标区域。研究了各潜在目标区域的多方向多级梯度特征,实现了单帧图像的小目标检测。研究结果表明,该方法能够有效抑制不均匀背景杂波,增强目标信号,提高单帧亮暗点目标的检测能力。对于信噪比为0.89的图像,可获得34.74的信噪比增益。

A grey level model of small targets in the spatial domain is established on the basis of thepoint spread theory in target imaging. The basic characteristics of the target, background and noise areanalyzed. After the open and close morphologic operation is implemented, the grey level variation ineach pixel position is derived. Thus, the potential target areas are determined. The multi-orientation andmulti-degree gradient of each potential area are studied. The detection of small targets is implementedin a single frame. The result shows that this method can effectively suppress uneven background clutter,enhance target signals and improve the detection of bright and dark small targets in a single frame. Foran image with a signal-to-noise ratio （SNR） of 0.89, a SNR gain of 34.74 can be obtained.